3916
Improvement of functional sensitivity using multi-echo EPI and BOLD response based HRF model at 7T1Center for Information and Neural Networks, National Institute of Information and Communications Technology, Suita, Japan, 2Graduate School of Frontier Biosciences, Osaka University, Suita, Japan
Synopsis
To study amygdala function, we used multi-echo EPI sequence as fMRI at 7T to reduce the effects of local magnetic field inhomogeneity in the region. Additionally, to increase sensitivity, we modified a hemodynamic response function model from an empirical model. These methods allowed us to detect robust activation of the amygdala and brain areas related to emotional face discrimination tasks in addition to increased functional connectivity between them. Using an appropriate acquisition method and statistical model can improve sensitivity in regions whose signal is suffered by local magnetic field inhomogeneity in ultra-high field fMRI studies.
Introduction
The quality of ultra-high field (UHF) fMRI (≥ 7 T) is affected by the inhomogeneity of local magnetic fields, which compromises signal drop and distort image, especially at the boundaries between air and brain tissue, such as the amygdala.1 To improve the signal in the regions, multi-echo (ME) sequence is an option that has been shown to increase functional sensitivity of combined echo images by using ME sequence.2 Another approach is to use a specific hemodynamic response model (HRF) based on blood oxygenation level-dependent (BOLD) responses to obtain an enhanced general linear model (GLM) of activation.3 In this study, using 7T multi-echo and multi-band echo planar imaging (MEMB-EPI) sequence and the HRF model based on the evoked response of the amygdala, we demonstrate robust activation of regions linked to emotional face processing and increased functional connectivity among these regions.Methods
Subjects: Seventeen volunteers with no history or evidence of neurological disorders participated in the study. Informed consent was obtained for all subjects prior to each study and all experiments were approved by the NICT.
Data acquisition: The experiments were performed on a 7T MRI scanner (Magnetom, Siemens Healthcare, Erlangen, Germany). Functional images were obtained at 2.5-mm isotropic resolution using MEMB-EPI sequence4 (TR = 1 s; three TEs = 12.0, 28.5, 42.3 ms; GRAPPA acceleration factor = 3; multi-band factor = 2, field of view = 220 mm, number of slices = 20, no gap). Anatomical images were acquired using magnetization-prepared rapid acquisition gradient echoes with 0.8-mm isotropic resolution of T1-weighted sagittal sections (flip angle = 5º, TR = 3100 ms, TE = 2.26 ms).
Study design: The face discrimination task detecting a fearful or happy face was conducted during the scan. For each trial, a face appeared on the screen for 2 sec. Participants were instructed to discriminate among faces, which were randomized across trials, by pressing the response key within a 1.5-sec time period. The total length of each trial was approximately 16 sec and 64 trials were shown per run.
Data analysis: A T2* weight-based summation scheme was applied to combine multiple echo images.2 Pre-processing was conducted on the combined images. This consisting of motion correction, geometric distortion correction, temporal filtering, and slice timing correction, was conducted on the combined images. These images were registered to MNI space and BOLD time series were extracted from the amygdala and averaged across all voxels. A new HRF was created based on the shape of the averaged time series. We applied a canonical HRF and the generated HRF models of GLM analysis to compare task-related brain activation. In addition, we defined common activated areas and extracted time series from those areas to calculate a correlation matrix between areas.
Results
Generated HRF model: The averaged time series of amygdala activation showed two negative peaks and one positive peak at 2-3, 6-7, and 11-12 seconds, respectively (Fig. 1a). We generated the HRF based on the shape of these averaged time series (Fig. 1b).
Common task-related brain areas: The generated HRF model showed robust activation in brain areas related to the task (Fig. 2). Nine common areas (bilateral temporal cortex, amygdala, hippocampus, pulvinar, and right inferior frontal gyrus) were identified in both hemispheres (Fig. 3).
Task-related functional connectivity: The functional connectivity of the generated HRF increased compared with that of the canonical HRF between common areas except for five areas (right temporal cortex-right hippocampus, right temporal cortex-left pulvinar, left temporal cortex-left pulvinar, left hippocampus-left pulvinar, and right amygdala-left amygdala), which showed slight decreases in connectivity (Fig. 4).
Discussion
Using MEMB-EPI and the generated HRF model derived from amygdala responses, we found robust brain activations related to emotional face processing and increased functional connectivity among brain regions suffered by local magnetic field inhomogeneity. All identified areas are closely related to emotional processing and their functional connectivity with the amygdala has been shown to be critical in a study on anxiety.5 In addition, the increased functional sensitivity from creating the HRF can explain that those brain areas have similar BOLD response time series. Since a previous study has found similar time series patterns in the amygdala,6 it is important to consider which specific HRF model is being used in such brain areas.7Conclusion
In this study, we suggest that the deliberate and appropriate choice of acquisition method and statistical model can maximize functional sensitivity in a UHF fMRI study. This approach should be considered for studies that require higher functional sensitivity due to significant signal loss or include multiple sub-regions in UHF fMRI.Acknowledgements
We thank Ichiro Fujimoto (BAIC), Tomoki Haji (NICT), and Hironori Nishimoto (NICT) for the data acquisition and Satoshi Tada (CiNet) for the experiment protocol. This study was partly supported by JSPS KAKENHI Grant Number JP18K07701 and JP18H04084 (IK).References
1. Sladky R, Baldinger P, Kranz GS, et al. High-resolution functional MRI of the human amygdala at 7 T. Eur J Radiol. 2013;82:728–733.
2. Poser BA, Norris DG. Investigating the benefits of multi-echo EPI for fMRI at 7 T. NeuroImage. 2009;45:1162–1172.
3. Handwerker DA, Ollinger JM, D’Esposito M. Variation of BOLD hemodynamic responses across subjects and brain regions and their effects on statistical analyses. NeuroImage. 2004;21:1639–1651.
4. Moeller S, Yacoub E, Olman CA, et al. Multiband multislice GE-EPI at 7 tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magn Reson Med. 2010;63:1144–1153.
5. Hakamata Y, Komi S, Moriguchi Y, et al. Amygdala-centred functional connectivity affects daily cortisol concentrations: a putative link with anxiety. Sci Rep 2017;7:8313.
6. Hrybouski S, Aghamohammadi-Sereshki A, Madan CR, et al. Amygdala subnuclei response and connectivity during emotional processing. Neuroimage. 2016;133:98-110.
7. Faull OK, Olivia K, Jenkinson M et al. Functional Subdivision of the human periaqueductal grey in respiratory control using 7tesla FMRI. NeuroImage. 2015;113:356–364.
Figures
